I Don't Understand the Push to Automate Assessments

Tales from an exhausted professor

Last week, I spent four days trying to get my head above water. Of course, I’m referring to the end-of-the-semester grading marathon.

I graded:

• Videos

• Poems

• Podcasts

• Screenplays

• Short Stories

• Reading Portfolios

• ChatGPT Transcripts

I’d spend hours a day cutting across different genres, depending on whether my students took an analytical approach, a creative approach, or something else.

I was exhausted.

Honestly, I’m still exhausted.

But it was worth every minute.

In reading and assessing my students’ work, I learned so much about how my students:

↳ Read
↳ Write
↳ Think
↳ Analyze
↳ Synthesize

I now know infinitely more about how my students process the world. And isn’t that the point?

The more I learn about them, the better I am able to support them. And I learn way more from going through the lengthy and something inefficient process of assessing their work than I ever would from an AI-written summary.

This is why I have a hard time understanding the push for automated grading.

Every week I get pitched new product that fits the bill. They will take over the time-consuming process of commenting on and grading student work (which, apparently, everyone hates) so that I can do more student-facing work.

Here is my response.

1. Assessing is student-facing work: Commenting is one of the main ways I interact with my students. They write back to me with questions. I respond. And so forth. The grade is meant to spark conversation.

2. I actually like assessing student work: No, I don’t like giving low marks or breaking the news when a student doesn’t put in the time or effort. But I love taking the time to look over my students’ creations carefully. And honestly, I don’t know if I would read my students’ work quite as closely if I weren’t the one commenting on them and assessing them.
   

Assessing doesn’t seem like something I should outsource.

It seems like something I should make more time for, by using AI to outsource other things.

What Worries Me

Not all innovation is equal.

There are innovations that make our lives easier, like automated grading. There are also innovations when we fundamentally rethink how we teach and connect with our students.

Neither is inherently bad.

But there are times when they brush up against each other.

Small, local solutions can prevent teachers and schools from making larger, systemic changes.

I think automated grading might be one of those things.

Instead of using AI to grade hundreds of submissions immediately, I choose to rethink assessment from the ground up.

It’s harder. But I think it’s more worth my time.

Some Caveats

I want to begin with a few clarifications, lest I’m greeted with a swarm of angry emails.

I understand the desire to automate grading, from a purely practical perspective. I also recognize that in some specific contexts, providing formative feedback to students through an AI system could be helpful.

I also fully recognize that there are ways to automate part of the process of actually writing the feedback, without handing evaluation over to a machine. For example, I’ve used a custom GPT to turn my notes on a project’s strengths and weaknesses into a comment. (I basically use the AI to fill out my own personalized template. It’s a fancy Excel sheet.) Other teachers might use AI to get feedback on their comments, using AI to make sure they’re getting their point across clearly.

What I’m still wrapping my head around are those programs (and there are many!) that grade a student submissions based on a provided rubric.

Even if I did it for a first pass, I don’t trust my ability to unsee what the AI assesses. If I take a student paper and run it through an AI program to get started, then I let AI frame my own feedback.